Study of Rota-Baxter Operators in Matrix C ∗ C^{*} -Algebras Motivated by Toeplitz Structures, and Applications to Sliding Mode Control

Marwa Ennaceur · 2026

Read Study of Rota-Baxter Operators in Matrix C ∗ C^{*} -Algebras Motivated by Toeplitz Structures, and Applications to Sliding Mode Control by Marwa Ennaceur (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-08126.txt.

Introduction

Marwa Ennaceur studies Rota-Baxter operators on the algebra of complex square matrices, with a heuristic nod to Toeplitz structures. The study classifies the operators that respect the C-star norm and works out the Lie brackets they induce, giving a clear structural picture of this operator family.

The algebra is then put to work in control theory. The classified operators deform the system matrices of discrete-time delayed systems under sliding mode control. Lyapunov-based matrix inequality conditions, rewritten in a tractable linear form, guarantee stability on the sliding surface and a bounded gain from uncertainty to state. The author is careful about the scope of each guarantee.

Catalog record

CREATORMarwa Ennaceur
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE92 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.08126

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Study of Rota-Baxter Operators in Matrix C ∗ C^{*} -Algebras Motivated by Toeplitz Structures, and Applications to Sliding Mode Control Marwa Ennaceur Abstract . This paper studies Rota-Baxter operators on the matrix C ∗ C^{*} -algebra M n ​ ( ℂ ) M_{n}(\mathbb{C}) , motivated by the discrete Toeplitz algebra (whose role is purely heuristic; see Remark 1.1 ). We provide a structural classification of such operators compatible with the C ∗ C^{*} -norm, analyze their induced Lie brackets, and apply them to deform system matrices in discrete-time delayed systems under sliding mode control. Lyapunov-based Bilinear Matrix Inequality conditions, together with a tractable linear reformulation via Q = X − 1 Q=X^{-1} , guarantee asymptotic stability on the sliding manifold and ℒ 2 \mathcal{L}_{2} -gain stability. The effective gain from uncertainty δ \delta to state x x is γ / μ \gamma/\sqrt{\mu} with μ = λ min ​ ( − ℳ ) > 0 \mu=\lambda_{\min}(-\mathcal{M})>0 determined a posteriori ; minimizing γ \gamma alone does not minimize this bound, which holds under zero extended initial conditions ( V 0 = 0 V_{0}=0 ). We work under the standing assumption m = n m=n (square actuation); a supplementary non-degenerate example with m = 1 m=1 , n = 2 n=2 illustrates LMI feasibility with Π ≠ 0 \Pi\neq 0 . All algebraic results are proved directly in M n ​ ( ℂ ) M_{n}(\mathbb{C}) ; no infinite-dimensional reduction is used. 2020 Mathematics Subject Classification 16W99, 17B99, 46L05, 93D05, 93C55 1. Introduction The discrete Heisenberg algebra arises as an associative algebra generated by elements u,u^{*},n subject to the relations [n,u]=-u,\quad[n,u^{*}]=u^{*},\quad[u,u^{*}]=p, where p is the rank-one projection onto the vacuum state. Equipped with the commutator [x,y]=xy-yx , this algebra inherits a Lie structure. Graded extensions yield Lie superalgebras, broadening the theoretical scope [ 11 , 2 ] . A faithful representation on the Hilbert space \ell^{2}(\mathbb{N}) identifies u with the unilateral shift U , u^{*} with its adjoint U^{*} , and n with the number operator Ne_{k}=ke_{k} . The norm closure of the generated algebra yields the discrete Toeplitz C^{*} -algebra [ 6 ] . While this infinite-dimensional setting provides a rich topological framework for spectral analysis and noncommutative deformations [ 5 , 1 ] , practical control synthesis requires finite-dimensional approximations. Specifically, we work with compressions of Toeplitz operators to the subspace \mathrm{span}\{e_{0},\ldots,e_{n-1}\}\subset\ell^{2}(\mathbb{N}) ; the finite-dimensional algebra M_{n}(\mathbb{C}) serves as the concrete representation algebra. Remark 1.1 (Scope of the Toeplitz motivation) . The discrete Toeplitz C^{*} -algebra motivates the structural setting of this paper but does not enter the proofs in an essential way, and its role is purely heuristic. The Rota-Baxter identity ( 1.1 ) is imposed directly on linear maps P:M_{n}(\mathbb{C})\to M_{n}(\mathbb{C}) and reduces to an explicit system of quadratic constraints on the matrix entries of P , verifiable by direct computation. A rigorous reduction from the infinite-dimensional Toeplitz algebra to M_{n}(\mathbb{C}) would require the invariance of the subspace \mathrm{span}\{e_{0},\ldots,e_{n-1}\} under all operators in the range of P . This condition fails already for the most natural generator: the unilateral shift U maps e_{n-1} to e_{n} , which lies outside \mathrm{span}\{e_{0},\ldots,e_{n-1}\} , so the subspace is not invariant under U . Consequently, no Rota-Baxter identity from the Toeplitz algebra compresses to one in M_{n}(\mathbb{C}) via this subspace, and any such reduction would additionally require spectral tools (Wiener–Hopf factorization, Fredholm index) that are not developed here. This extension is deferred to future work. All algebraic results in Sections 3 – 6 are proved entirely within M_{n}(\mathbb{C}) ; the Toeplitz algebra is cited only as the source of structural intuition. Within this setting, Rota-Baxter operators play a central role. A linear map P:\mathcal{A}\to\mathcal{A} satisfies the Rota-Baxter identity of weight \lambda\in\mathbb{R} : (1.1) P(x)P(y)=P(xP(y))+P(P(x)y)+\lambda P(xy),\quad\forall\,x,y\in\mathcal{A}. Originally introduced to formalize integration and summation [ 10 ] , Rota-Baxter operators also generate solutions to classical and modified Yang-Baxter equations, thereby linking associative and Lie structures [ 8 ] . In operator algebras, they induce algebraic deformations that can model cumulative effects, filtering, or memory-dependent interactions. In control theory, sliding mode control (SMC) is renowned for its robustness against matched uncertainties and time delays [ 16 , 7 ] . We consider discrete-time delayed systems of the form (1.2) \begin{cases}x(k+1)=Ax(k)+A_{d}x(k-\tau)+Bu(k)+D\delta(k),\\ s(k)=Cx(k),\\ u(k)=-Kx(k)-\rho\,\sat\bigl(s(k)/\phi\bigr),\end{cases} where x(k)\in\mathbb{R}^{n} , u(k)\in\mathbb{R}^{m} , B\in\mathbb{R}^{n\times m} , D\in\mathbb{R}^{n\times p} , \tau\in\mathbb{N} is a known delay, \delta(k)\in\mathbb{R}^{p} is a bounded uncertainty satisfying \|\delta(k)\|_{2}\leq\delta_{\max} for all k , and \sat(\cdot) denotes the componentwise saturation function \sat(v/\phi)_{i}=\begin{cases}v_{i}/\phi&\text{if }|v_{i}|\leq\phi,\\ \mathrm{sgn}(v_{i})&\text{if }|v_{i}|>\phi,\end{cases} with boundary layer width \phi>0 to mitigate chattering in discrete time. We assume throughout that m=n (the square case) and p=m , i.e., C\in\mathbb{R}^{m\times n} is a square matrix. This ensures that CB_{P}\in M_{m}(\mathbb{C}) is a square matrix, and its invertibility (Assumption 4.1 ) is well-defined in the ordinary sense without recourse to pseudoinverses. The hypothesis p=m is adopted for notational convenience: it ensures that the congruence transformation \Sigma=\mathrm{diag}(Q,Q,I_{p},I_{n}) in Remark 5.7 produces a well-structured LMI with consistent block dimensions. Physically, p=m means the number of disturbance channels equals the number of control inputs, a structural simplification that can be relaxed at the cost of a more complex block structure in ( 5.7 ). The case m<n (rectangular B ) or p<m (fat output map C ) requires a separate treatment involving either right-inverses or a reformulation of the equivalent control, and is deferred to future work. The supplementary non-degenerate numerical example of Section 7.2 relaxes this assumption to m=1 , n=2 for purely illustrative purposes. Assumption 1.2 (Bounded initial conditions) . The initial state and initial delay segment are bounded: \sup_{k\geq 0}\|x(k)\|\leq r_{0}<\infty and \sup_{k\geq 0}\|x(k-\tau)\|\leq r_{d,0}<\infty prior to closed-loop activation. This holds whenever the open-loop system is BIBO stable and the initial data x(0) and \{x(-\tau),\ldots,x(-1)\} are finite. In practice, r_{0} and r_{d,0} are set conservatively from the known initial conditions; they are treated as design parameters in the reaching analysis of Section 4 . Important note on the design order. The bound r_{0} is fixed from open-loop data before the control gains K and \rho are selected. However, the validity of r_{0} as a uniform bound throughout the closed-loop reaching phase is a condition that is verified a posteriori after K is chosen, via condition ( 4.5 ). There is therefore no strict a priori non-circularity: one must first fix r_{0} and \rho_{\max} from open-loop data, then choose K satisfying ( 4.4 ), and only then verify ( 4.5 ). If the condition fails, r_{0} must be increased (conservatively) or K must be redesigned. The full sequential procedure is spelled out in Remark 4.7 . When the open-loop system is not BIBO stable, this assumption must be imposed explicitly as a constraint on the initial data. All operator norms appearing in the SMC analysis (Sections 4 – 5 ) are spectral (operator) norms \|\cdot\|_{\mathrm{op}}=\sigma_{\max}(\cdot) , unless otherwise stated. The Frobenius norm \|\cdot\|_{F} appears only in the structural analysis of P_{+} in Section 3 (Example 3.5 (b) and Remark 3.9 ), where it is always labeled explicitly. A non-standard norm requirement. Throughout the SMC analysis, we impose condition ( 4.4 ): \|A_{P}-B_{P}K\|_{\mathrm{op}}<1 . This is strictly stronger than Schur stability (spectral radius \rho(A_{P}-B_{P}K)<1 ), and is not automatically satisfied even when A_{P}-B_{P}K is Schur stable.

Read the complete text